The present invention relates to amplitude qualification of peaks in a signal peak detection system, and, more particularly, to the setting of a reference level for the amplitude qualification.
In order to reconstruct binary data recorded upon a magnetic disc, the polarity-reversing magnetic flux peaks must be detected and converted into digital pulses. One method of eliminating false data due to spurious peaks or noise is to require that any peak be of a minimum amplitude in order to be considered as a possible data peak. Such a system is shown in FIG. 1 and disclosed in U.S. Pat. No. 4,081,756 to Price, et al.
Referring to FIG. 1, a disc read head 12 converts magnetic flux reversals into a signal which is processed through a preamplifier 14 and an equalizer 16. An automatic gain control amplifier 18 adjusts the gain in response to changes in signal level. The signal is then processed through a differentiator 20, the output of which is connected to a zero crossing detector 22. The zero crossings will correspond to peaks of the detected signal. In a separate channel, the signal is provided to a pair of comparators 24, 26 which receive a reference input from a potentiometer 28. The output of comparators 24, 26 is provided through a series of logic gates 30 and to an AND gate 32. Thus, any signal which exceeds, in either the positive or the negative direction, the threshold set by potentiometer 28 will produce a logical one at one input of AND gate 32. The other input of AND gate 32 is an output of zero crossing detector 22 indicating a peak. This output is delayed by a delay circuit 34 so that the two channel paths are equal. The output of AND gate 32 thus provides an indication of a peak only if the peak exceeds an amplitude set by potentiometer 28.
One problem with the use of amplitude qualification is that the signal level may vary from point to point on a magnetic disc. For instance, near the center of the disc where the bits are packed more closely together, the signal level will be smaller. One method of compensating for this is to make the amplitude threshold a percentage of the signal level. This method is shown in U.S. Pat. No. 4,001,884 to Herbst. Such a system, however, relies upon an RC circuit to follow the signal level, and thus cannot instantaneously adjust to changes in signal level.